Obremenitveno testiranje spletnih aplikacij v javnih oblakih

Ivanšek, Simon (author), Ciglarič, Mojca (mentor)

V magistrski nalogi se ukvarjamo s problemom testiranja skalabilnosti in stroškov povezanih s povečanim obiskom, s katerim se danes srečuje vse več podjetij. Pri testiranju se osredotočimo na javna oblaka Amazon Web Services in Google Compute Platform. K reševanju problema pristopimo z obremenitvenim testiranjem tako, da razvijemo vzorčno aplikacijo in orodja, s katerimi jo obremenitveno testiramo. Vzorčno aplikacijo in orodja razvijemo po standardu TPC-W, ki ga uporabimo tudi za izvajanje obremenitvenega testa. Po zgledu metrik iz standarda TPC-W definiramo dve novi metriki, s katerima merimo in vrednotimo skalabilnost in stroške, povezane s povečanim obiskom. Meritve izvedemo najprej z ročnim skaliranjem, nato še z vklopljenim avtomatskim skaliranjem. Rezultate meritev ovrednotimo z definiranimi metrikami, izdelamo stroškovni model za omenjena oblaka in ju med seboj primerjamo.

• Language: Unknown
• Keywords: obremenitveno testiranje, računalništvo v oblaku, generiranje bremen
• Work type: Master's thesis/paper
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2015
• PID: 20.500.12556/RUL-72729
• COBISS.SI-ID: 1536575683
• Publication date in RUL: 29.09.2015

Secondary language

• Language: Unknown
• Title: Stress testing of web applications in public clouds

Abstract

In this master thesis we are solving the problem of scalability testing and estimating the costs associated with increased load on web application, with which many companies are facing nowadays. We focus on testing of Amazon Web Services and Google Cloud Platform public clouds. Our approach to solving the problem is with load testing, that's why we develop sample web application and tools to execute the load tests. For developing the web application and tools we used TPC-W standard. We also execute the load test according to TPC-W standard. Based on TPC-W metrics, we develop two new metrics that we use for evaluating load tests. First we perform the measurements with manual scaling and then with automated scaling. After that we evaluate the measurements with newly defined metrics, make the cost model for each cloud and compare them.

• Keywords: stress testing, cloud computing, load generation